Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Dynamic molecular switches with hysteretic negative differential conductance emulating synaptic behaviour

Abstract

To realize molecular-scale electrical operations beyond the von Neumann bottleneck, new types of multifunctional switches are needed that mimic self-learning or neuromorphic computing by dynamically toggling between multiple operations that depend on their past. Here, we report a molecule that switches from high to low conductance states with massive negative memristive behaviour that depends on the drive speed and number of past switching events, with all the measurements fully modelled using atomistic and analytical models. This dynamic molecular switch emulates synaptic behavior and Pavlovian learning, all within a 2.4-nm-thick layer that is three orders of magnitude thinner than a neuronal synapse. The dynamic molecular switch provides all the fundamental logic gates necessary for deep learning because of its time-domain and voltage-dependent plasticity. The synapse-mimicking multifunctional dynamic molecular switch represents an adaptable molecular-scale hardware operable in solid-state devices, and opens a pathway to simplify dynamic complex electrical operations encoded within a single ultracompact component.

This is a preview of subscription content, access via your institution

Access options

Buy article

Get time limited or full article access on ReadCube.

$32.00

All prices are NET prices.

Fig. 1: The dynamic molecular junction.
Fig. 2: Mechanism of PCET.
Fig. 3: Electrical characteristics of the Au-S-C10-HATNA//Ga2O3/EGaIn junctions.
Fig. 4: Model of the dynamic switch.
Fig. 5: Synaptic function.
Fig. 6: Demonstration of Pavlovian learning and Boolean logic gates using time plasticity in the DMS.

Data availability

Data are available in the supplementary materials and on https://dataverse.harvard.edu/privateurl.xhtml?token=efd1a016-cc24-41ab-891b-5217fa6dd56d. Source data are provided with this paper.

Code availability

Codes are available from Harvard Dataverse at https://dataverse.harvard.edu/privateurl.xhtml?token=efd1a016-cc24-41ab-891b-5217fa6dd56d.

References

  1. Van De Burgt, Y., Melianas, A., Keene, S. T., Malliaras, G. & Salleo, A. Organic electronics for neuromorphic computing. Nat. Electron. 1, 386–397 (2018).

    Article  Google Scholar 

  2. Zhu, J., Zhang, T., Yang, Y. & Huang, R. A comprehensive review on emerging artificial neuromorphic devices. Appl. Phys. Rev. 7, 011312 (2020).

    Article  CAS  Google Scholar 

  3. Upadhyay, N. K. et al. Emerging memory devices for neuromorphic computing. Adv. Mater. Technol. 4, 1800589 (2019).

    Article  Google Scholar 

  4. Chen, T. et al. Classification with a disordered dopant-atom network in silicon. Nature 577, 341–345 (2020).

    Article  CAS  Google Scholar 

  5. Keene, S. T. et al. A biohybrid synapse with neurotransmitter-mediated plasticity. Nat. Mater. 19, 969–973 (2020).

    Article  CAS  Google Scholar 

  6. Oh, S., Hwang, H. & Yoo, I. K. Ferroelectric materials for neuromorphic computing. APL Mater. 7, 91109 (2019).

    Article  Google Scholar 

  7. Wang, L., Lu, S. R. & Wen, J. Recent advances on neuromorphic systems using phase-change materials. Nanoscale Res. Lett. 12, 1–22 (2017).

    Article  Google Scholar 

  8. Wan, Q., Sharbati, M. T., Erickson, J. R., Du, Y. & Xiong, F. Emerging artificial synaptic devices for neuromorphic computing. Adv. Mater. Technol. 4, 1900037 (2019).

    Article  Google Scholar 

  9. Xu, W., Min, S. Y., Hwang, H. & Lee, T. W. Organic core–sheath nanowire artificial synapses with femtojoule energy consumption. Sci. Adv. 2, e1501326 (2016).

    Article  Google Scholar 

  10. Ratera, I. & Veciana, J. Playing with organic radicals as building blocks for functional molecular materials. Chem. Soc. Rev. 41, 303–349 (2012).

    Article  CAS  Google Scholar 

  11. Klajn, R. Spiropyran-based dynamic materials. Chem. Soc. Rev. 43, 148–184 (2014).

    Article  CAS  Google Scholar 

  12. Bléger, D. & Hecht, S. Visible-light-activated molecular switches. Angew. Chem. Int. Ed. 54, 11338–11349 (2015).

    Article  Google Scholar 

  13. Sorrenti, A., Leira-Iglesias, J., Markvoort, A. J., De Greef, T. F. A. & Hermans, T. M. Non-equilibrium supramolecular polymerization. Chem. Soc. Rev. 46, 5476–5490 (2017).

    Article  CAS  Google Scholar 

  14. Van Rossum, S. A. P., Tena-Solsona, M., Van Esch, J. H., Eelkema, R. & Boekhoven, J. Dissipative out-of-equilibrium assembly of man-made supramolecular materials. Chem. Soc. Rev. 46, 5519–5535 (2017).

    Article  Google Scholar 

  15. Bear, M. F., Connors, B. W. & Paradiso, M. A. Neuroscience: Exploring the Brain (Wolters Kluwer, 2016).

  16. Eccles, J. C. & McIntyre, A. K. Plasticity of mammalian monosynaptic reflexes. Nature 167, 466–468 (1951).

    Article  Google Scholar 

  17. Segura, J. L., Juárez, R., Ramos, M. & Seoane, C. Hexaazatriphenylene (HAT) derivatives: from synthesis to molecular design, self-organization and device applications. Chem. Soc. Rev. 44, 6850–6885 (2015).

    Article  CAS  Google Scholar 

  18. Wang, R. et al. Cyclic and normal pulse voltammetric studies of 2,3,6,7,10,11-hexaphenylhexazatriphenylene using a benzonitrile thin layer-coated glassy carbon electrode. J. Phys. Chem. B 107, 9452–9458 (2003).

    Article  CAS  Google Scholar 

  19. Perrin, M. L. et al. Large negative differential conductance in single-molecule break junctions. Nat. Nanotechnol. 9, 830–834 (2014).

    Article  CAS  Google Scholar 

  20. Migliore, A. & Nitzan, A. Irreversibility and hysteresis in redox molecular conduction junctions. J. Am. Chem. Soc. 135, 9420–9432 (2013).

    Article  CAS  Google Scholar 

  21. Schwarz, F. et al. Field-induced conductance switching by charge-state alternation in organometallic single-molecule junctions. Nat. Nanotechnol. 11, 170–176 (2016).

    Article  CAS  Google Scholar 

  22. Garrigues, A. R. et al. A single-level tunnel model to account for electrical transport through single molecule- and self-assembled monolayer-based junctions. Sci. Rep. 6, 26517 (2016).

    Article  CAS  Google Scholar 

  23. Warren, J. J. & Mayer, J. M. in Proton-Coupled Electron Transfer: A Carrefour of Chemical Reactivity Traditions (eds Formosinho, S. & Barroso, M.) 1–31 (The Royal Society of Chemistry, 2012).

  24. Migliore, A., Polizzi, N. F., Therien, M. J. & Beratan, D. N. Biochemistry and theory of proton-coupled electron transfer. Chem. Rev. 114, 3381–3465 (2014).

    Article  CAS  Google Scholar 

  25. Mayer, J. M. Understanding hydrogen atom transfer: from bond strengths to Marcus theory. Acc. Chem. Res. 44, 36–46 (2011).

    Article  CAS  Google Scholar 

  26. Kim, S. et al. Experimental demonstration of a second-order memristor and its ability to biorealistically implement synaptic plasticity. Nano Lett. 15, 2203–2211 (2015).

    Article  CAS  Google Scholar 

  27. Wang, Z. et al. Memristors with diffusive dynamics as synaptic emulators for neuromorphic computing. Nat. Mater. 16, 101–108 (2017).

    Article  CAS  Google Scholar 

  28. Thompson, R. in International Encyclopedia of the Social & Behavioral Sciences (eds Smelser, N. J. & Baltes, P. B.) 6458–6462 (Pergamon, 2001).

  29. Van De Burgt, Y. et al. A non-volatile organic electrochemical device as a low-voltage artificial synapse for neuromorphic computing. Nat. Mater. 16, 414–418 (2017).

    Article  Google Scholar 

  30. Ruiz Euler, H. C. et al. A deep-learning approach to realizing functionality in nanoelectronic devices. Nat. Nanotechnol. 15, 992–998 (2020).

    Article  CAS  Google Scholar 

  31. Kathan, M. et al. Light-driven molecular trap enables bidirectional manipulation of dynamic covalent systems. Nat. Chem. 10, 1031–1036 (2018).

    Article  CAS  Google Scholar 

  32. Chakma, P. & Konkolewicz, D. Dynamic covalent bonds in polymeric materials. Angew. Chem. Int. Ed. 58, 9682–9695 (2019).

    Article  CAS  Google Scholar 

  33. Cafferty, B. J. et al. Robustness, entrainment, and hybridization in dissipative molecular networks, and the origin of life. J. Am. Chem. Soc. 141, 8289–8295 (2019).

    Article  CAS  Google Scholar 

  34. Ashkenasy, G., Hermans, T. M., Otto, S. & Taylor, A. F. Systems chemistry. Chem. Soc. Rev. 46, 2543–2554 (2017).

    Article  CAS  Google Scholar 

  35. Shi, J. et al. The influence of water on the charge transport through self-assembled monolayers junctions fabricated by EGaIn technique. Electrochim. Acta 398, 139304 (2021).

    Article  CAS  Google Scholar 

  36. Ai, Y. et al. In-place modulation of rectification in tunneling junctions comprising self-assembled monolayers. Nano Lett. 18, 7552–7559 (2018).

    Article  CAS  Google Scholar 

  37. Barber, J. R. et al. Influence of environment on the measurement of rates of charge transport across AgTS/SAM//Ga2O3/EGaIn junctions. Chem. Mater. 26, 3938–3947 (2014).

    Article  CAS  Google Scholar 

  38. Han, Y. & Nijhuis, C. A. Functional redox-active molecular tunnel junctions. Chem. Asian J. 15, 3752–3770 (2020).

    Article  CAS  Google Scholar 

  39. Han, Y. et al. Electric-field-driven dual-functional molecular switches in tunnel junctions. Nat. Mater. 19, 843–848 (2020).

    Article  CAS  Google Scholar 

  40. Egger, D. A., Liu, Z. F., Neaton, J. B. & Kronik, L. Reliable energy level alignment at physisorbed molecule–metal interfaces from density functional theory. Nano Lett. 15, 2448–2455 (2015).

    Article  CAS  Google Scholar 

  41. Larsen, C. B. et al. Synthesis and optical properties of unsymmetrically substituted triarylamine hexaazatrinaphthalenes. Eur. J. Org. Chem. 2017, 2432–2440 (2017).

    Article  CAS  Google Scholar 

  42. Chen, X. et al. Molecular diodes with rectification ratios exceeding 105 driven by electrostatic interactions. Nat. Nanotechnol. 12, 797–803 (2017).

    Article  CAS  Google Scholar 

  43. Yuan, L., Jiang, L., Thompson, D. & Nijhuis, C. A. On the remarkable role of surface topography of the bottom electrodes in blocking leakage currents in molecular diodes. J. Am. Chem. Soc. 136, 6554–6557 (2014).

    Article  CAS  Google Scholar 

  44. Yuan, L., Breuer, R., Jiang, L., Schmittel, M. & Nijhuis, C. A. A molecular diode with a statistically robust rectification ratio of three orders of magnitude. Nano Lett. 15, 5506–5512 (2015).

    Article  CAS  Google Scholar 

  45. Reus, W. F. et al. Statistical tools for analyzing measurements of charge transport. J. Phys. Chem. C 116, 6714–6733 (2012).

    Article  CAS  Google Scholar 

  46. Yuan, L. et al. Transition from direct to inverted charge transport Marcus regions in molecular junctions via molecular orbital gating. Nat. Nanotechnol. 13, 322–329 (2018).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank the Ministry of Education (MOE, awards no. MOE2018-T2-1-088 and no. MOE2019-T2-1-137) and the Prime Minister’s Office, Singapore, under its Medium Sized Centre program for supporting this research. D.T. acknowledges support from Science Foundation Ireland (SFI) under awards no. 15/CDA/3491 and no. 12/RC/2275_P2 and supercomputing resources at the SFI/Higher Education Authority Irish Center for High-End Computing (ICHEC). E.d.B. and C.N. acknowledge support from the US National Science Foundation (grant no. ECCS#1916874). D.Q. acknowledges the support of the Australian Research Council (grant no. FT160100207). C.H. and B.B. gratefully acknowledge funding from the Deutsche Forschungsgemeinschaft (German Research Foundation) Project-ID 433682494-SFB 1459.

Author information

Authors and Affiliations

Authors

Contributions

C.A.N. conceived and supervised the project. D.T. conducted the atomistic modelling. E.d.B. and C.N. conducted the numerical modelling. Y.W. and A.B. synthesized the compounds, Y.W. and Y.H. performed the CV and ultraviolet–visible measurements. Z.Z. and D.-C.Q. performed the angle-resolved XPS and near-edge X-ray absorption fine structure measurements and analysis. Y.W. and Q.Z. performed the IV electrical measurements. H.P.A.G.A. designed, performed and analysed the synaptic emulation measurements. S.S. and F.A.A. conducted the millisecond-scale pulse experiments. C.H. and B.B. conducted and analysed the sum-generation frequency experiments. A.L. developed the Origin code for electrical data analysis. C.A.N., E.d.B., D.T. and H.P.A.G.A. wrote the manuscript and all the authors commented on it.

Corresponding authors

Correspondence to Enrique del Barco, Damien Thompson or Christian A. Nijhuis.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Materials thanks Ferdinand Grozema, Tae-Woo Lee and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 The periodic DFT calculations.

a, The calculated PDOS of Au-bound S-C10-HATNA with the nitrogen sites plotted as a thick black line. The full molecule is shown as a dashed grey line and contributions from carbon, hydrogen, oxygen, and sulphur sites are plotted as green, red, purple, and blue lines, respectively. b, The calculated PDOS of post-PCET Au-S-C10-(H6-HATNA). (c) Computed surfaces of the empty electron-acceptor LUMO → LUMO + 2 levels in Au-S-C10-HATNA together with the mid-gap state created at +1.1 eV in the first protonation step generating Au-S-C10-H+-HATNA. d, The calculated PDOS on the molecule of HATNA SAMs on Au(111) for all oxidation and protonated states.

Source data

Extended Data Fig. 2 Measurement of potentiation as a function of negative pulse amplitude as shown in Fig. 5b.

a-b, I(V) profile of the junction used in the measurement in linear (a) and logarithmic scale (b). c-e, pulse sequence used in the measurement with current (c), logarithm of current (d) and voltage (e) as a function of time, with colour used to help distinguish sections of the sequence. A first negative voltage pulse of amplitude Vp (with tp = 1 s) was applied, a delay at 0 V was set for td = 1 s, and then a second negative voltage pulse of magnitude Vp was applied, followed by a series of positive voltage pulses of amplitude Vr = 1.25 V to reset the junction. This sequence was repeated each time varying the amplitude of the negative voltage pulses Vp from 0.5 V to 2 V and back. After half a sequence to train the junction, the current amplitude of the second negative pulse was compared to that of the first negative pulse (using the first data point of each pulse) to determine the potentiation as a function of Vp. Time t = 0 corresponds to the time the measurement was started, immediately after junction formation.

Source data

Extended Data Fig. 3 Measurement of potentiation as a function of negative pulse duration as shown in Fig. 5c.

a-b, I(V) profile of the junction used in the measurement in linear (a) and logarithmic scale (b). c-e, pulse sequence used in the measurement with current (c), logarithm of current (d) and voltage (e) as a function of time, with colour used to help distinguish sections of the sequence. A first negative voltage pulse of amplitude Vp = -2V and duration tp was applied, a delay at 0 V was set for td = 1 s, a second negative voltage pulse of magnitude Vp was applied, followed by a series of positive voltage pulses of amplitude Vr = 1.25 V to reset the junction. This sequence was repeated each time varying Vp the amplitude of the negative voltage pulses from 0.5 V to 2 V and back, and after half a sequence to train the junction, the current amplitude of the second negative pulse was compared to that of the first negative pulse to determine the potentiation as a function of Vp. Time t = 0 corresponds to the time the measurement was started, immediately after junction formation.

Source data

Extended Data Fig. 4 Measurement of potentiation as a function of inter-pulse delay td as shown in Fig. 5d.

a-b, I(V) profile of the junction used in the measurement in linear (a) and logarithmic scale (b). c-e, pulse sequence used in the measurement with current (c), logarithm of current (d) and voltage (e) as a function of time, with colour used to help distinguish sections of the sequence. A first negative pulse was applied, followed by a delay of duration td at 1.25 V, then a second pulse, and finally a positive pulse is applied to reset the junction. Each pulse pair was used to extract two data points: one with td > 0 and one with td < 0, whereby a positive (negative) value of td signifies that the reference pulse occurs before (after) the measured pulse. The relative potentiation was measured as the ratio of the difference between the current of the measured pulse and the current of the reference pulse (here using the values of current in the middle of the pulses) to the current of the reference pulse. Time t = 0 corresponds to the time the measurement was started, immediately after junction formation.

Source data

Extended Data Fig. 5 Measurement of potentiation as a function of positive pulse duration as shown in Fig. 5e.

a-b, I(V) profile of the junction used in the measurement in linear (a) and logarithmic scale (b). c-e, pulse sequence used in the measurement with current (c), logarithm of current (d) and voltage (e) as a function of time, where colour is used to help distinguish sections of the sequence. A negative reference voltage pulse (Vp = -1.5 V, tp = 1 s) was applied, a positive voltage pulse of varying amplitude (duration td = 2 s) was applied, and a second negative voltage pulse (Vp = -1.5 V, tp = 1 s) was applied. This sequence was repeated each time varying Vr the amplitude of the positive voltage pulse from 0.5 V to 2 V and back, and after one training sequence (~500 s), the current amplitude of the second negative pulse was compared to that of the reference negative pulse to determine the potentiation as a function of Vr. Time t = 0 corresponds to the time the measurement was started, immediately after junction formation.

Source data

Extended Data Fig. 6 Voltage cycling measurements for the junctions.

Voltage cycling showing 100 000 voltage pulses with amplitudes of -2 V and +1 V and pulse time of 20 µs (black) and current response (red); b, zoomed-in version of the voltage cycling data in a to show the pulse parameters and current response more clearly; c, Measurement of potentiation as a function of short positive voltage pulses of +1 V amplitude and 2 ms pulse time period (black), and corresponding current response of the junction (red) showing PPD. Due to the basic nature of HATNA, HATNA can be protonated representing the on-state (that is, the (HATNA)H+ state as explained in the main text and shown in Fig. 2f-g (and experimentally proven with vibrational sum-frequency generation spectroscopy of HATNA monolayers described in Supplementary Section S3.3). A series of short positive voltages deprotonates HATNA turning it off which is observed as reduction in the current response, as seen here. More experimental details on the pulse parameters are given in Supplementary Information Section S5.3.

Source data

Extended Data Fig. 7 Measurement demonstrating the Pavlov experiment (Hebbian learning).

a-b, I(V) profile of the junction used in the measurement in linear (a) and logarithmic scale (b). c, pulse sequence showing Hebbian learning (from Fig. 6b-e). The e-ANN works as follows: two source-measure units represent two artificial neurons N1 and N2 delivering voltage inputs in separate channels feeding into a third channel (see Fig. 6a) N3 where the current received is the output. N1 is connected via a fixed resistor (1kΩ) to N3 and gives the output unconditionally. N2 includes the junction in the off state such that pulses from N2 do not initially result in a substantial signal in N3. It corresponds to ‘dog hears bell’. However, firing pulses in N1 and N2 results in a PPF process in N2, ‘training’ it to deliver an output in N3 when pulses are fired from N2 thereafter. The sequences of voltage pulses are as follows. N1 alternates between pulses at -1 V and -1.1 V, each lasting 2 s, repeated 5 times. N2 alternates between pulses at +1.5 V and -1.1 V, each lasting 2 s, repeated 5 times. When pulses are applied from N1 and N2 simultaneously, the voltage across the dynamic switch can be estimated to be alternating between +2.4 V and -0.1 V approximately, according to Kirchhoff’s circuit laws, which excites the junction (enhances its conductance) and thus the coupling to N3.

Source data

Extended Data Fig. 8 Demonstration of remaining logic gates.

a, XOR; b, NAND; c, NOR; d, NXOR, as per the operation system described in the main text and Fig. 6, with current and voltage traces and truth tables. More details on the pulse parameters are given in Supplementary Information Section S5.2.3.

Source data

Supplementary information

Source data

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wang, Y., Zhang, Q., Astier, H.P.A.G. et al. Dynamic molecular switches with hysteretic negative differential conductance emulating synaptic behaviour. Nat. Mater. 21, 1403–1411 (2022). https://doi.org/10.1038/s41563-022-01402-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41563-022-01402-2

This article is cited by

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing